Trust Wallet CEO AI Cyberattack Warning Raises Concerns About Banking Security
The Trust Wallet CEO AI cyberattack warning has drawn attention to an emerging cybersecurity challenge: hackers are increasingly using artificial intelligence to automate attacks against businesses and financial institutions.
On November 17, Trust Wallet CEO Felix Fan shared his concerns about the ability of traditional banks and corporations to defend themselves against autonomous AI-powered threats.
In a post on X, Fan suggested that many companies remain unprepared for attacks that can operate with limited human supervision.
He expressed particular concern about traditional banks and corporations that hold substantial amounts of customer funds while relying on IT departments structured around conventional working hours.
Fan also suggested that he personally felt more comfortable keeping funds in Trust Wallet than in traditional banks under these circumstances.
That statement reflects his personal assessment, not independent evidence that Trust Wallet is safer than regulated banking institutions.
His comments followed a November 16 report from Google's Threat Intelligence Group (GTIG), which documented how cybercriminals and state-linked threat actors are adopting increasingly sophisticated AI systems.
The report describes attackers using autonomous frameworks to identify vulnerabilities, conduct reconnaissance, develop malicious code, and coordinate different stages of cyber intrusions.
Rather than relying on a human operator to manually perform every action, these systems can complete multiple tasks through coordinated AI agents.
The development raises important questions about whether conventional cybersecurity teams can respond quickly enough when attackers operate at machine speed.
What Did Trust Wallet CEO Felix Fan Say About AI Cyberattacks?
Fan's warning focused on the gap between the speed of modern cyberattacks and the operational structure of many established businesses.
His argument was that financial institutions holding customer assets could become attractive targets for increasingly autonomous cybercriminal operations.
Traditional banks generally maintain dedicated cybersecurity departments, monitoring systems, and incident response procedures.
However, Fan questioned whether their existing organizational structures are adequately prepared for attacks that can continuously discover and exploit weaknesses.
AI-powered tools can operate around the clock.
They can repeatedly examine software, search for vulnerable systems, and adjust their approach based on automated feedback.
This potentially allows attackers to conduct activities that previously required teams of skilled human operators.
Fan's comments also highlight a broader debate about financial custody.
Banks typically hold customer deposits and provide centralized account management.
Self-custody cryptocurrency wallets, including Trust Wallet, allow users to control their own private keys.
These arrangements involve different security models.
A bank may provide fraud investigation processes, account restrictions, and certain regulatory protections.
A self-custody wallet gives the owner direct control over assets but generally cannot reverse a valid blockchain transaction if private keys are compromised.
Therefore, Fan's preference for self-custody should not be interpreted as proof that cryptocurrency wallets eliminate cyberattack risks.
Google Confirms AI-Powered Cyberattacks Are Becoming More Autonomous
The timing of Fan's warning is significant because Google published new threat intelligence findings just one day earlier.
On November 16, Google Threat Intelligence Group and Mandiant released research examining the increasing use of autonomous AI systems in cybercrime and espionage.
The report found that attackers are moving beyond using chatbots simply to generate malicious text or assist with coding.
Some threat actors are now deploying coordinated AI frameworks capable of executing multiple stages of an intrusion.
Google identified tools including ARTEX, CAIRN, CyberStrikeAI, and PentAGI.
These frameworks can support activities such as vulnerability discovery, reconnaissance, exploit development, and automated execution.
Google also observed real-world deployment of certain frameworks.
For example, researchers tracked a threat actor using a customized PentAGI-based system to conduct reconnaissance and vulnerability scanning against organizations in Asia between late August and early October 2026.
The findings suggest that AI is increasingly becoming part of operational cyberattack infrastructure rather than remaining an experimental technology.
However, the research does not establish that all cyberattacks are now fully autonomous.
Many operations still involve human direction, and the level of automation varies significantly between different threat actors.
How Autonomous AI Cyberattacks Work
Traditional cyberattacks often require human operators to perform a sequence of tasks.
An attacker might search for vulnerable systems, examine software configurations, develop an exploit, gain access, and then move through the compromised network.
Each stage can require specialized knowledge and considerable time.
AI agents can help automate portions of this process.
For example, an autonomous framework may analyze publicly accessible systems, identify potential vulnerabilities, and test whether a known weakness can be exploited.
It may then use feedback from those tests to determine its next action.
Some frameworks coordinate multiple specialized agents.
One agent might gather information about a target system.
Another could analyze software behavior.
A third might evaluate whether a vulnerability is exploitable.
The agents can exchange information and adjust their actions based on the results.
This approach can reduce the time required to perform complex operations.
Google's research found that attackers are also using AI to analyze software patches.
When a company releases a security update, attackers can compare the patched version with the previous version to identify the vulnerability that was fixed.
AI tools can accelerate this process and help develop exploits targeting organizations that have not yet installed the update.
This creates a shorter window for defenders to protect vulnerable systems.
The danger is not that AI has invented an entirely new category of cybercrime.
It is that existing attack techniques can potentially become faster, cheaper, and easier to scale.
Why Traditional Banks Could Become Targets for Autonomous AI Attacks
Banks manage highly sensitive financial infrastructure.
Their systems may contain customer information, payment instructions, account records, authentication credentials, and connections to other financial institutions.
This makes them attractive targets for financially motivated criminals.
AI-powered attacks could increase pressure on several areas of banking security.
Continuous vulnerability scanning
Attackers can use automated tools to repeatedly examine internet-facing systems for known weaknesses.
AI may help prioritize vulnerabilities and accelerate the development of exploitation techniques.
Employee credential theft
Financial institutions depend on employees, administrators, and service providers with varying levels of system access.
Attackers may use AI-generated phishing messages, impersonation techniques, or other social engineering methods to target those individuals.
Third-party software risks
Banks frequently rely on external technology providers.
A weakness in a software supplier or service provider can create an indirect route into sensitive infrastructure.
Faster movement after initial compromise
Once attackers gain access, automated tools may help identify additional systems and opportunities to escalate privileges.
However, these risks do not mean banks are universally unprepared.
Many large financial institutions already operate continuous security monitoring, fraud detection, incident response teams, and automated controls.
Fan's comments represent a warning about preparedness, not a verified assessment that banking cybersecurity teams work only during standard office hours.
Google's findings establish that autonomous attack capabilities are advancing, but they do not demonstrate that traditional banks are collectively less secure than cryptocurrency wallets.
Google Identifies AI Systems That Can Exploit Vulnerabilities Without Human Intervention
One of the most concerning findings in Google's report involves a framework identified as exploit_agent.
Researchers discovered an exposed staging server hosting an AI-powered vulnerability research system.
The framework was configured to compare software versions, analyze security patches, identify potential vulnerabilities, and generate exploits.
Google also identified components capable of automatically executing those exploits without requiring a human operator to approve each step.
The system integrated multiple AI models and supported additional post-compromise activities.
These included internal reconnaissance and attempts to establish further access within affected environments.
The research suggests that some attackers are beginning to automate tasks that previously required significant manual effort.
Google also documented AI-assisted attacks involving known vulnerabilities in PaperCut print management software.
Security researchers demonstrated how quickly working exploits could be developed following disclosure of the vulnerabilities.
Separately, Google noted that AI-generated exploit claims are not always reliable.
Some automatically produced proof-of-concept code fails to work.
This distinction matters because AI can accelerate attack development, but it does not guarantee successful exploitation.
Nevertheless, the overall trend creates additional pressure on organizations to deploy security patches quickly and improve continuous monitoring.
Why Cryptocurrency Companies Are Also Vulnerable to AI-Powered Attacks
Although Fan expressed greater confidence in self-custody, cryptocurrency businesses are not immune to AI-assisted cybercrime.
Google's November report identified a suspected North Korean threat cluster, tracked as MIDNIGHT NEPTUNE, targeting cryptocurrency organizations in North America and Europe.
The attackers reportedly created fraudulent recruiter profiles using AI-generated images and directed employees toward malicious meeting platforms.
After compromising targeted environments, they attempted to exploit internal development tools and software configurations.
In one example, attackers modified Claude Code configuration files so that a malicious backdoor could be downloaded when developers used the coding assistant.
The findings demonstrate that AI-enabled attacks can target both traditional financial institutions and cryptocurrency businesses.
The wider crypto industry has already experienced major security incidents.
Bitnxt previously reported on the Binance $6.3 billion hack, in which attackers compromised exchange infrastructure and transferred customer-related assets.
That incident should not be described as a confirmed autonomous AI attack. It illustrates the broader risks associated with privileged access and centralized financial infrastructure.
Another example involves the SlowMist iOS crypto wallet attack warning, which described malicious tools capable of searching compromised devices for cryptocurrency recovery phrases.
Together, these developments show that neither traditional banks nor cryptocurrency platforms can rely on a single security mechanism.
Is Trust Wallet Really Safer Than Traditional Banks?
Fan's comments raise a question that deserves careful consideration.
Trust Wallet is a self-custody cryptocurrency wallet.
Users generally control the private keys needed to authorize transactions from their wallets.
This differs from conventional banking, where customers typically rely on a financial institution to hold deposits, maintain accounts, and process transactions.
Self-custody reduces dependence on a centralized institution for controlling blockchain assets.
However, it also places substantial responsibility on the wallet owner.
If an attacker steals a recovery phrase, the attacker may be able to transfer the associated cryptocurrency without permission.
Blockchain transactions are generally irreversible once finalized.
Traditional banks, meanwhile, may provide fraud monitoring, transaction restrictions, dispute procedures, and applicable deposit insurance protections.
These protections vary by institution, account type, and jurisdiction.
A compromised bank account and a compromised self-custody wallet can therefore produce very different recovery outcomes.
The appropriate comparison is not simply which technology is more modern.
It is which risks each custody model creates and which protections are available when something goes wrong.
Fan's statement should be understood as an expression of concern about centralized financial infrastructure, rather than an independently established ranking of financial security.
AI Cybersecurity Could Become an Arms Race Between Hackers and Financial Institutions
Google's research highlights a broader shift in cybersecurity.
Attackers are using AI to automate activities that previously required substantial manual effort.
Defenders are also adopting AI-powered systems to identify suspicious activity and respond more quickly.
Google said its security operations already include specialized detection mechanisms for certain autonomous attack frameworks.
Mandiant has also developed defensive guidance intended to help organizations counter AI-assisted intrusions.
These efforts are important because human analysts may struggle to respond manually to every event generated by automated attacks.
However, AI-powered defense introduces its own challenges.
Automated systems can produce false positives, miss unfamiliar threats, or make incorrect decisions.
Financial institutions must therefore balance speed with reliability and appropriate human oversight.
The most effective approach is likely to combine automated detection, strong access controls, continuous monitoring, timely patching, and well-tested incident response procedures.
For banks, cryptocurrency exchanges, and wallet providers, the challenge is increasingly about responding to attacks before they can spread through critical infrastructure.
Bitnxt Take: AI Is Changing Cybersecurity, but Self-Custody Is Not a Complete Solution
At Bitnxt, we believe Felix Fan's warning highlights a genuine change in the cybersecurity landscape.
The most concerning development is not simply that criminals are using AI.
Cybercriminals have used automation for years.
What makes the current generation of tools different is their growing ability to coordinate tasks, interpret feedback, and adapt their actions without constant human direction.
Google's findings suggest that some of these capabilities are already being used in real attacks.
That creates a serious challenge for financial institutions operating large and complex technology systems.
However, we think Fan's comparison between banks and self-custody wallets requires more context.
Traditional banks can be attractive targets because they manage large amounts of money and sensitive customer information.
But they also operate within established regulatory frameworks and often maintain extensive security infrastructure.
Self-custody wallets reduce reliance on financial intermediaries, yet they create different risks.
A user who loses control of a recovery phrase may have no practical way to reverse an unauthorized transaction.
We therefore do not believe the emergence of autonomous AI attacks automatically makes self-custody safer than banking.
The more important issue is whether financial institutions and cryptocurrency companies are adapting their defenses quickly enough.
Another concern is the increasing accessibility of sophisticated attack tools.
AI frameworks may allow less experienced attackers to automate activities that previously required specialized cybersecurity expertise.
This could increase the number of organizations facing attempted intrusions.
We also believe the cryptocurrency industry should pay particular attention to AI-powered social engineering.
Attackers can use convincing fake identities, fraudulent recruitment messages, and malicious software development tools to compromise employees.
Google's findings involving cryptocurrency companies demonstrate that these risks are already relevant to the industry.
Bitnxt's view is that autonomous AI cyberattacks will make continuous security monitoring, stronger access controls, and faster incident response essential for both traditional finance and cryptocurrency platforms.
The future of financial security will depend less on whether assets are held by a bank or a blockchain wallet and more on how effectively the entire system protects users against evolving threats.
What Should Banks and Crypto Companies Do to Prepare?
Organizations facing AI-powered cyber threats need to strengthen security across several areas.
First, they should prioritize timely software updates and vulnerability management.
Google's findings show that attackers can use AI to analyze patches and develop exploits against systems that remain unprotected.
Second, companies should restrict privileged access to sensitive systems.
Employee accounts, administrative credentials, and development tools can become valuable targets.
Third, continuous security monitoring is increasingly important.
Organizations need mechanisms capable of detecting suspicious behavior outside conventional working hours.
Fourth, businesses should review third-party software and AI tools.
Google's research demonstrates that attackers may exploit development environments and software supply chains rather than directly targeting customer-facing systems.
Finally, incident response procedures should be tested regularly.
Automated detection is valuable, but organizations also need clear processes for containing attacks, preserving evidence, and restoring affected services.
For individual cryptocurrency users, protecting recovery phrases, keeping devices updated, and verifying transaction requests remain essential precautions.
What Happens Next for Autonomous AI Cybersecurity?
The latest Google report suggests that AI-assisted cyberattacks will continue evolving.
Researchers expect threat actors to improve the autonomy and sophistication of their tools.
As these systems become more capable, the time between vulnerability discovery and attempted exploitation may shrink further.
Financial institutions will need to evaluate whether existing cybersecurity processes can respond quickly enough.
Cryptocurrency companies face similar challenges, particularly because blockchain transactions may be difficult or impossible to reverse.
The next major developments will likely involve improved AI-powered threat detection, stronger security controls around autonomous agents, and greater scrutiny of AI tools integrated into financial infrastructure.
However, the available evidence does not establish that a particular bank has suffered a successful autonomous AI attack as a result of the vulnerabilities discussed by Fan.
His warning is best understood as a call for improved preparedness rather than confirmation of a specific banking breach.
Conclusion
Trust Wallet CEO Felix Fan has warned that traditional banks and corporations may be unprepared for the growing threat of autonomous AI cyberattacks.
His comments follow Google Threat Intelligence research documenting how attackers are using AI frameworks to automate vulnerability research, exploitation, and other stages of cyber intrusions.
The findings demonstrate that AI-powered cybercrime is moving beyond simple chatbot-assisted attacks toward more coordinated and automated operations.
However, self-custody cryptocurrency wallets are not immune to cybersecurity threats, and the evidence does not establish that they are universally safer than regulated banking institutions.
For Bitnxt, the central lesson is that financial security must evolve alongside artificial intelligence.
Both traditional banks and cryptocurrency companies will need stronger monitoring, improved operational controls, and faster incident response capabilities to protect users in an increasingly automated threat environment.













































